Effects of orlistat vs. metformin on weight loss-related clinical variables in women with PCOS: systematic review and meta-analysis

Search strategy and study selection

The following databases were searched for prospective studies and randomised clinical trials (RCTs) published until May 2015: Medline, Cochrane Central Register of Controlled Trials, and LILACS. No limits were set on publication date or language. This systematic review was registered in PROSPERO (http://www.crd.york.ac.uk/PROSPERO/) under number CRD42014012877. The following basic search strategy was developed for PubMed and modified as needed for other databases: “Polycystic Ovary Syndrome” or “Ovary Syndrome, Polycystic” or “Syndrome, Polycystic Ovary” or “Stein-Leventhal Syndrome” or “Stein-Leventhal Syndrome” or “Syndrome, Stein-Leventhal” or “Sclerocystic Ovarian Degeneration” or “Ovarian Degeneration, Sclerocystic” or “Sclerocystic Ovary Syndrome” or “Polycystic Ovarian Syndrome” or “Ovarian Syndrome, Polycystic” or “Sclerocystic Ovaries” or “Ovary, Sclerocystic” or “Sclerocystic Ovary” and tetrahydrolipstatin or THLP or 1-((3-hexyl-4-oxo-2-oxetanyl)methyl)dodecyl-2-formamido-4-methylvalerate or tetrahydrolipstatin or Xenical or “Roche brand of orlistat” or “Hoffmann-La Roche brand of orlistat” or Alli or “GlaxoSmithKline brand of orlistat.”

The selection criteria for the studies were as follows: diagnosis of PCOS using Rotterdam 31 or NIH criteria 32, intervention with any dose of orlistat for at least 8 weeks, and comparison of orlistat with placebo or metformin or any anti-obesity drug. Regarding the studies comparing orlistat with anti-obesity drugs other than metformin in PCOS, or comparing the use of orlistat in PCOS women vs. healthy controls, only the results from the PCOS/orlistat arm were considered in the present systematic review. Studies with lifestyle interventions associated with pharmacological treatment were also included in the systematic review if all the groups in the study received the same intervention.

Primary outcomes were changes in body mass index (BMI), weight, waist circumference, insulin resistance markers and total testosterone levels following orlistat treatment. Secondary outcomes were changes in lipid profile and menstrual cyclicity, ovulation rate and ovarian morphology at ultrasound and adverse effects.

In addition, the reference lists of identified studies were searched. The most complete study was chosen to avoid duplication if the same patient populations were reported in several publications. Whenever necessary, authors were contacted to obtain additional data from published materials.

Data extraction and quality control assessment

Two reviewers (SKG and FMM) independently screened titles/abstracts for selection of articles for full text review. Disagreements were resolved by a third reviewer (PMS) or consensus discussion. The full text of selected articles was independently reviewed by the two initial reviewers. Cochrane Collaboration tools for assessing the risk of bias in randomised trials 33 were also independently applied by two reviewers (SKG and FMM), with disagreements resolved by a third reviewer (PMS) or consensus discussion. The following information was extracted from studies: name of first author, publication year, country, PCOS diagnostic criteria, type of study, intervention, number of subjects in each arm, age, length of study and lifestyle intervention.

Statistical analysis

Data are presented as mean (±SEM) at baseline and after treatment. The comparison between baseline and after-treatment data within each arm (orlistat or metformin) is presented as mean percentage change from baseline (±SEM). Data from RCTs comparing the effects of orlistat vs. metformin were combined by fixed effects model meta-analysis. Variables of interest expressed as mean (±SEM) were included in the meta-analysis if they were present in at least two studies with the same unit of measurement. Mean percentage changes from baseline achieved with orlistat or metformin were recorded for each variable/outcome, and the difference (orlistat minus metformin) between these percentage changes was considered as the effect size. Therefore, results are presented as mean differences with 95% confidence intervals (95% CI). A p-value < 0.05 was considered statistically significant. Heterogeneity was assessed using the I² statistics and Cochran's Q test. It was not possible to meta-analyse the effect of orlistat vs. placebo on clinical variables associated with weight loss because we found only one study comparing orlistat with placebo. All analyses were conducted using the Meta package from R software version 3.0.

Flow chart of study selection

Figure 1 provides details of the study selection. The primary search identified 3951 articles. After title and abstract screening, 14 potentially eligible studies were retrieved for full text review. Of these 14 articles, five were excluded: one did not meet the inclusion criteria and four overlapped with other studies. Therefore, nine studies were included in the systematic review: three prospective, non-randomised studies comparing the use of orlistat in PCOS women and healthy controls 34-36 and six RCTs (four studies comparing orlistat with metformin in PCOS women, one comparing orlistat with sibutramine in PCOS women, and one comparing orlistat with placebo in PCOS women) 37-42.

Characteristics of included studies

Table 1 summarises the characteristics of the nine studies included in the systematic review. Two studies focused on Caucasian women from the UK 37, 38, two on Iranian women 39, 41, one on Indian women 42 and four on Greek women 34-36, 40. Eight of these studies employed Rotterdam criteria for diagnosis of PCOS and one used NIH criteria [hyperandrogenemia (free androgen index > 8) and history of oligomenorrhoea (cycle length, < 21 days or > 35 days; < 8 cycles per year) or amenorrhoea and hirsutism (Ferriman-Gallwey score >8) 37.

All nine studies included overweight/obese PCOS women. In four, orlistat was compared with metformin 37-39, 42. One study compared orlistat and sibutramine 40. In that study, a normal weight PCOS group receiving metformin for 6 months was also included. Another study compared orlistat and placebo 41. Finally, three studies compared orlistat in overweight/obese PCOS women vs. overweight/obese healthy controls 34-36.

Four studies (comparing orlistat vs. metformin or vs. sibutramine) 37-40 did not report the mean age for each group separately. Only the mean age for the overall participants was presented, and that ranged from 25.7 to 27.0 years. One study 42 included women younger than 40 years of age. In the study comparing orlistat vs. placebo 41, the mean age was 26.8 ± 5.2 in PCOS women treated with orlistat and 27.4 ± 3.3 in the placebo group. In the studies comparing PCOS and healthy controls 34-36, the age of PCOS women ranged from 25.4 to 26.1 years, vs. 30.6 to 32.1 years for controls, and study duration ranged from 12 to 24 weeks. Two 24-month studies also reported 12-week data.

The four RCTs included in the systematic review, comparing orlistat vs. metformin in PCOS women 37-39, 42, lasted 12 weeks. Three studies used the same dosages for both drugs 37-39 and one used a lower dosage of orlistat 42. One of these studies also offered general dietary guidance 37, while another provided guidance on low-calorie diet and lifestyle modification 42. The sample size ranged from 10 to 40 women for both the PCOS group treated with orlistat and the PCOS group treated with metformin.

One study compared orlistat with sibutramine in overweight/obese women with PCOS 40. These women also received low-calorie diet guidance and instructions for performing moderate intensity aerobic exercise for at least 3 h/week. In that study, an additional group of lean women with PCOS received metformin for 6 months. Another study comparing orlistat vs. metformin was excluded, because baseline and final results were not available 42.

Only one RCT had a placebo control group 41. Participants were instructed to follow a diet (1200–1800 kcal) and were encouraged to walk for 30 min daily.

Three additional studies included in the systematic review compared the use of orlistat in overweight/obese PCOS women and overweight/obese healthy controls 34-36. All three studies offered an energy-restricted diet and one recommended moderate intensity aerobic exercise for at least 3 h/week 35.

The risk of bias of randomised trials included in the systematic review is shown in Figure 2.

Qualitative data synthesis

Quantitative data synthesis and meta-analysis

A meta-analysis was performed to compare the effects of orlistat vs. metformin on BMI, HOMA, testosterone and insulin levels (Figure 3). Fixed effects models were used because heterogeneity was not significant. The main results obtained in the three studies included in this meta-analysis 37-39 are presented in Table 2. One of the four RCTs comparing orlistat vs. metformin in PCOS women was excluded because data on baseline and final results were not available 42.

It was not possible to meta-analyse the effect of orlistat vs. placebo on clinical variables associated with weight loss because we found only one study comparing orlistat with placebo. Similarly, a meta-analysis of orlistat effects described in three prospective, non-controlled studies (before and after-treatment comparisons) 34-36 and three RCTs 40-42 was not performed because neither the measure of variability (% change) nor baseline and after-treatment values were available.

Systematic review

Our systematic review revealed that orlistat was associated with BMI reduction after 12 (ranging from 4.48% to 8.10%) and 24 weeks (12.9%) in women with PCOS. The XENDOS study, a prospective clinical trial with obese patients from the general population, found that orlistat (120 mg three times daily) produced significantly higher weight loss as compared with placebo in 4 years (5.2% vs. 2.8%) 25.

Currently, guidelines for the treatment of obesity in the general population recommend an initial weight loss of 5–10% within 6 months 43. Moderate weight loss can produce health benefits, including improved glycaemic control and lipid profile and reduced risk of DM2 44-47. Studies have shown improvement in menstrual cyclicity and a higher spontaneous ovulation rate and pregnancy in obese women with PCOS following a 5% decrease in body weight 48-54.

Regarding the effects of orlistat on markers of IR, five of eight studies showed a positive response, while three others found no significant improvement in these parameters 37, 41, 42. A difference in treatment duration may, however, help explain this discrepancy: the two studies that did not find significant changes in IR markers lasted only 12 weeks, whereas those with longer duration observed a significant effect of orlistat. In addition, the lack of statistical significance despite a marked reduction in fasting insulin (12.5%) and HOMA (10.8%) in the study of Jayagopal et al. 37 might have resulted from the small sample size, only 10 patients. In turn, Diamanti-Kandarakis et al. 34 showed that there were significant improvements in IR markers with orlistat regardless of weight loss only in the PCOS group.

In the present review, seven of eight studies 34-37, 39, 41, 42 showed a decrease in testosterone levels in association with orlistat administration. Even though a reduction in BMI and IR markers was observed in these studies, the decrease in testosterone levels reported by Diamanti-Kandarakis et al. 34 cannot be attributed exclusively to weight reduction, because it persisted even after adjustment for changes in BMI.

None of the studies we analysed was able to demonstrate any improvement in menstrual regularity, ovarian volume or number of follicles with the use of orlistat. However, longer duration studies are needed to investigate the benefits of weight reduction, improved insulin resistance and androgen levels to restore or ameliorate menstrual cyclicity.

Among the five studies describing the effect of intervention with orlistat on lipid profile 35, 37, 39, 41, 42, only one was unable to show a significant beneficial effect 37, possibly because of the small sample size. The prospective XENDOS study observed that orlistat led to significantly greater improvement than placebo in total cholesterol and LDL-cholesterol after 4 years of treatment in the general population 25.

The main side effects described for orlistat were related to the drug's mechanism of action of decreasing fat absorption from the intestinal lumen – oily stool, flatulence and diarrhoea with faecal urgency occurring in around 5–40% of participants 34, 37, 39, 41. In contrast, although not directly assessed in the present review, the side effects of metformin (nausea, mild abdominal pain and diarrhoea) are known to be dose-dependent 22, 37, 39.

Meta-analysis

Weight loss is considered the first-line treatment for obese PCOS women. Anti-obesity drugs have also been considered for these women. Orlistat promotes weight loss by partially preventing intestinal fat absorption 55, 56. Metformin improves insulin action and has been considered a second-line treatment for obese and/or insulin-resistant PCOS women 19-24. Thus, the present meta-analysis was conducted to analyse the evidence for the differential effects of orlistat vs. metformin treatment on weight loss-associated clinical variables in overweight/obese women with PCOS.

The results of the present meta-analysis show that orlistat was comparable to metformin in reducing BMI. A recently published meta-analysis with more than 11,000 overweight/obese individuals from the general population showed that the use of metformin together with lifestyle change promoted a weight reduction of 1.92 kg (2.94–0.89; p = 0.11), while orlistat plus lifestyle change reduced weight by 3.05 kg (3.75–2.35; p = 0.0001); however, when the two drugs were compared, the difference was not statistically significant (p = 0.07) 57.

We did not observe a difference between metformin and orlistat in terms of insulin and HOMA reduction. Merlotti et al. 58 recently reported an effect of various interventions in reducing the risk of DM2 in obese individuals. All strategies assessed in that meta-analysis reduced the risk of developing DM2. Similar effectiveness was found for physical activity and diet [OR 0.44 (0.36–0.52)], antidiabetic drugs (metformin, glitazones, glinides, alpha-glucosidase inhibitors) [OR 0.53 (0.33–0.86)] and weight loss-promoting drugs (orlistat, bezafibrate, phentermine/topiramate controlled release) [OR 0.52 (0.35–0.78)].

Evidence indicates that weight reduction in PCOS improves hyperlipidaemia, reduces IR and increases SHBG concentration, thereby reducing biochemical hyperandrogenism and improving menstrual cyclicity 54, 59, 60. In the present meta-analysis, the lack of difference between the effect of metformin or orlistat on testosterone levels might be related to the lack of difference between these drugs in terms of their effect on weight loss.

We observed significant heterogeneity in the analysis of BMI. This heterogeneity limits the interpretation of data and suggests similar effects of orlistat and metformin. The analyses of HOMA, insulin and testosterone levels revealed very low heterogeneity between studies. In turn, as a result of the limited number of studies, we were not able to determine whether there was publication bias.

One strength of this systematic review and meta-analysis is that all studies considered the Rotterdam criteria to diagnose PCOS, with the exception of one study, which employed NIH criteria. Therefore, the PCOS population was certainly homogeneous. Limitations of this meta-analysis are the reduced number of studies, the small sample sizes and the absence of studies with long-term interventions. However, similar analyses are not available in the literature, and this study represents the first evidence for orlistat effects in women with PCOS.